diff --git a/optimum_benchmark/trackers/latency.py b/optimum_benchmark/trackers/latency.py
index 01a3c7a1..0555578f 100644
--- a/optimum_benchmark/trackers/latency.py
+++ b/optimum_benchmark/trackers/latency.py
@@ -275,7 +275,7 @@ def get_latency(self) -> Latency:
             ]
         else:
             latencies = [
-                (start_event - end_event) for start_event, end_event in zip(self.start_events, self.end_events)
+                (end_event - start_event) for start_event, end_event in zip(self.start_events, self.end_events)
             ]
 
         assert all(latency >= 0 for latency in latencies)
@@ -357,7 +357,7 @@ def track(self):
         self.per_token_start_events.extend(self.per_token_events[:-1])
         self.per_token_end_events.extend(self.per_token_events[1:])
 
-    def __call__(self, *args, scores: torch.FloatTensor, **kwargs):
+    def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor):
         if self.is_pytorch_cuda:
             event = torch.cuda.Event(enable_timing=True)
             event.record()
@@ -383,7 +383,7 @@ def get_prefill_latency(self) -> Latency:
             ]
         else:
             latencies = [
-                (start_event - end_event)
+                (end_event - start_event)
                 for start_event, end_event in zip(self.prefill_start_events, self.prefill_end_events)
             ]
 
@@ -401,7 +401,7 @@ def get_decode_latency(self) -> Latency:
             ]
         else:
             latencies = [
-                (start_event - end_event)
+                (end_event - start_event)
                 for start_event, end_event in zip(self.decode_start_events, self.decode_end_events)
             ]
 
@@ -419,7 +419,7 @@ def get_per_token_latency(self) -> Latency:
             ]
         else:
             latencies = [
-                (start_event - end_event)
+                (end_event - start_event)
                 for start_event, end_event in zip(self.per_token_start_events, self.per_token_end_events)
             ]
 
@@ -516,7 +516,7 @@ def get_step_latency(self) -> Latency:
             ]
         else:
             latencies = [
-                (start_event - end_event)
+                (end_event - start_event)
                 for start_event, end_event in zip(self.per_step_start_events, self.per_step_end_events)
             ]
 
@@ -534,7 +534,7 @@ def get_call_latency(self) -> Latency:
             ]
         else:
             latencies = [
-                (start_event - end_event)
+                (end_event - start_event)
                 for start_event, end_event in zip(self.call_start_events, self.call_end_events)
             ]
 
@@ -583,7 +583,7 @@ def get_latency(self) -> Latency:
             ]
         else:
             latencies = [
-                (start_event - end_event) for start_event, end_event in zip(self.start_events, self.end_events)
+                (end_event - start_event) for start_event, end_event in zip(self.start_events, self.end_events)
             ]
 
         assert all(latency >= 0 for latency in latencies)